Skip to content
NimbusVPS
Looking glass

Measure it yourself before you pay us anything.

Latency claims on a marketing page are worth nothing next to a traceroute you ran yourself, so here are the targets and the commands. Read the notice below first: the addresses are examples until each site opens, and we are not going to let you spend an afternoon pinging something that was never there.

These targets are examples, not live endpoints
Example addresses reserved for documentation (RFC 5737). They do not answer — the live per-site targets go up when each site opens to orders. The commands underneath each one are correct and worth keeping; the addresses they point at simply have nothing behind them yet. The previous set were live addresses that belonged to other networks, which is a worse problem than an unhelpful page.

Amsterdam, Netherlands

Nikhef / AMS-IX campus · 100 Gbps

Details →
test ipv4 198.51.100.10
$ ping -c 10 198.51.100.10
$ mtr --report --report-cycles 20 198.51.100.10
$ curl -o /dev/null https://ams.nimbusvps.com/1000MB.test
from Frankfurt
6 ms
from London
8 ms
from Paris
10 ms
from New York
78 ms

Bucharest, Romania

M247 Bucharest · 40 Gbps

Details →
test ipv4 198.51.100.20
$ ping -c 10 198.51.100.20
$ mtr --report --report-cycles 20 198.51.100.20
$ curl -o /dev/null https://buh.nimbusvps.com/1000MB.test
from Frankfurt
26 ms
from Istanbul
22 ms
from Amsterdam
31 ms
from Moscow
42 ms

Reykjavík, Iceland

Advania Thor · 20 Gbps

Details →
test ipv4 198.51.100.30
$ ping -c 10 198.51.100.30
$ mtr --report --report-cycles 20 198.51.100.30
$ curl -o /dev/null https://rkv.nimbusvps.com/1000MB.test
from London
21 ms
from Amsterdam
27 ms
from New York
46 ms
from Frankfurt
32 ms

Sofia, Bulgaria

Telepoint Sofia · 40 Gbps

Details →
test ipv4 198.51.100.40
$ ping -c 10 198.51.100.40
$ mtr --report --report-cycles 20 198.51.100.40
$ curl -o /dev/null https://sof.nimbusvps.com/1000MB.test
from Frankfurt
30 ms
from Bucharest
9 ms
from Istanbul
17 ms
from Amsterdam
36 ms

Chișinău, Moldova

Trabia MD · 20 Gbps

Details →
test ipv4 198.51.100.50
$ ping -c 10 198.51.100.50
$ mtr --report --report-cycles 20 198.51.100.50
$ curl -o /dev/null https://kiv.nimbusvps.com/1000MB.test
from Bucharest
11 ms
from Frankfurt
38 ms
from Kyiv
19 ms
from Amsterdam
44 ms

Panama City, Panama

Panama Digital Gateway · 20 Gbps

Details →
test ipv4 198.51.100.60
$ ping -c 10 198.51.100.60
$ mtr --report --report-cycles 20 198.51.100.60
$ curl -o /dev/null https://pty.nimbusvps.com/1000MB.test
from Miami
42 ms
from Bogotá
28 ms
from New York
71 ms
from São Paulo
118 ms

Victoria, Seychelles

NimbusVPS Cage SC-1 · 10 Gbps

Details →
test ipv4 198.51.100.70
$ ping -c 10 198.51.100.70
$ mtr --report --report-cycles 20 198.51.100.70
$ curl -o /dev/null https://sez.nimbusvps.com/1000MB.test
from Dubai
48 ms
from Mumbai
61 ms
from Singapore
92 ms
from Frankfurt
141 ms

Kansas City, United States

Wholesale Data Center KC1 · 100 Gbps

Details →
test ipv4 198.51.100.80
$ ping -c 10 198.51.100.80
$ mtr --report --report-cycles 20 198.51.100.80
$ curl -o /dev/null https://mci.nimbusvps.com/1000MB.test
from Chicago
12 ms
from Dallas
14 ms
from New York
28 ms
from Los Angeles
41 ms
Measuring properly

Four tests that tell you something.

A single ping tells you almost nothing. These take about five minutes and give you a real answer.

Latency and jitter

ping -c 50 198.51.100.10

Fifty packets, not four. Read the mdev figure at the end — jitter matters more than mean latency for anything interactive, and a low mean with high jitter is worse than a slightly higher stable path.

Route quality

mtr --report --report-cycles 50 198.51.100.10

Shows loss at every hop. Loss at an intermediate hop with none at the destination is usually ICMP rate-limiting, not a real problem. Loss that persists to the final hop is real.

Throughput

curl -o /dev/null -w "%{speed_download}\n" https://ams.nimbusvps.com/1000MB.test

A single stream is limited by your own TCP window and by any congested hop in between. Run it three times at different hours before drawing conclusions.

Multi-stream throughput

aria2c -x8 -s8 -d /tmp https://ams.nimbusvps.com/1000MB.test

Eight parallel streams. If this is dramatically faster than the single-stream result, the bottleneck is per-connection window scaling rather than raw capacity.

Test files are rate-limited, not throttled
Test file downloads are capped at 1 Gbps per source address to stop the endpoints being used as a free CDN. That is the same port speed a production server gets, so the number you measure is the number you would see — but please do not loop the download.

Prefer to test from inside the network?

Deploy the $2.90 plan for a month, run whatever benchmarks you like from the box itself, and if the numbers do not satisfy you take the 7-day refund. That is a more honest test than anything we could publish, and it costs less than a coffee.

Deploy a test serverNetwork specifications

Deploy in the next five minutes.

Pick a location, size the box, pay in crypto. No account signup wall, no ID, no waiting on a human.

Choose your specsSee plans & pricing

7-day money-back guarantee · No KYC · Cancel any time from the panel